Help file for running XPREP and/or SADABS

(March, 2007)


Once the data has been integrated by SAINT, there will be a series of data files in the (smartuser)(apexuser)/mydir subdirectory in which you are operating. These need to be further converted before you can use them with either TEXSAN or SHELXTL. In the process we will also apply corrections for crystal decay (which is very rare in the way that we usually take data) and for absorption (which is common, even on weakly absorbing crystals, due to the diffraction geometry which we use). We also want to get some information about our data set and crystal; space group, density, statistics, etc. There are two programs which we use for this purpose, XPREP and SADABS. For the purposes of this exercise the raw filenames are filename0.raw, filename1.raw, filename2.raw, filename3.raw, and filenamem.raw.

Normally we will run XPREP to determine space group, density and absorption coefficient and to test the effects of absorption. It is then recommended that you use SADABS on the data set to apply a different type of absorption correction and also to allow it to do the other things that it does to the data. SADABS is a release version of the program, but much of what it does is mysterious. It is significantly better than XPREP in cases of high absorption or other problems, but not much different in well-behaved cases. If you have really serious absorption you will need to do a face-indexed absorption correction.

1) XPREP

XPREP is a prgram developed by George Sheldrick.  It is designed to perform space group analyses, apply absorption corrections, and examine your data in a number of useful ways (Laue patterns, reflection measurements at differnt resolution shells and so-on). There is documentation of older versions included in the SHELXTL suite of software. It reads either raw data files from SAINT or hkl files from other programs such as SADABS. It also reads P4P files to get information about the crystal. It writes a file with extension PRP (i.e. filenamem.prp), which is a printable output file, and optionally writes other hkl files. What follows here is an abbreviated description. We start the program as follows.

For the purposes of these examples, we will use "smartuse" to define the user/instrument on which you collected data.  It can be substituted with apexuser if you collected data on the APEX instrument.

<...smartuse/mydir> % xprep filenamem <CR>

XPREP assumes either an hkl or raw extension to the file name.

If you forget to give the filename, and just type xprep, XPREP will ask for a filename. You have to specify the extension. Type: filenamem.raw.  If you have had to enter the filename separately you will then be prompted for which type of data format the file is in select: "[4] SHELX HKLF 4 format or SAINT ._rf file".

The program will read the raw data file, and also the file filenamem.p4p to get information about the cell. In the interaction which follows, there will always be a default value for a response which will be shown by the program in square brackets, [xx].  To select the default response, just hit <CR>. However, you do not always want the default response.

Be sure to read what the display says and react accordingly.

XPREP will show you displays which are designed to allow you and it to choose the space group. You should be aware that XPREP is very conservative about choosing a space group, and that it misses even P21/c on a regular basis. You may very well need to force it to take the correct space group.

The following is a step-by-step analysis of each window and what is going on within XPREP.

First you will see an analysis of the data to determine the lattice type.  Usually the default (highlighted in brackets) is acceptable.

You will then be prompted for a search for Higher Metric Symmetry [H], followed by output for the reduced cell parameters.  You will often be given several options for what the Laue group is.  Select the most appropriate (again usually the default).

You will then begin a space group analysis [S].  The lattice type is re-determined and you will be shown a list of systematic absences.  Examine these and determine if XPREP has obtained the correct space group.  Often there will be several choices.  use your knowledge and best judgement to determine if the correct space group has been selected.  If not, select it yourself.  Be aware of any potential cell transformations.  You may need to lie to the program for purposes of absorption correction.

XPREP will then perform a low-level data analysis [D].  Accept the defaults but be sure to examine the analysis of the data at various resolution shells, for completeness, R(int) and R(sym).  You have to go through this process or XPREP will return here until this routine is performed.

You will return to the main menu and be prompted to input the Cell Contents [C].  enter this routine and fill in the molecular formula as best you are able.  Note: XPREP utilises some common naming conventions for various groups (Me = methyl, Et = ethyl etc).  You can also enter molecules separated into moieties by enclosing them in parentheses.  This allows input of multiple of the same type of moiety (if present in your molecules) by including the number of such groups AFTER the parentheses.  XPREP will calculate mu, the absorption coefficient.  Note the value down somewhere so that you can calculate mu*r or mu*t for the absorptioin corrections.

WARNING: In order to do an absorption correction correctly, the cell must NOT be transformed before the correction is applied. Frequently the correct choice of space group will result in an un-announced transformation of the unit cell. Check the Matrix section of the output periodically to make sure that the cell has not been transformed without your consent. It may be necessary to lie to the program about your space group, at least initially, to get it to leave your cell alone. (Note that this lie does absolutely no damage to your data, since TeXsan will re-determine your space group and appropriate cell transformation when it processes the data.)

After you have entered the cell contents, the next step is to check absorption using the [A] option at the main menu. The official "type" of absorption correction is Psi Scan [P].   You will then be prompted with:

Psi and raw data MUST be indexed on original cell !
 <Enter> to abort now, 'C' to continue:

Press C <CR> to continue the absorption correction process.

You will then be prompted for the wavelength at which the data were collected.  The default is correct for data colleted on our laboratory instruments.

The tolerance for direction cosines should be set to 0.05, the default of 0.01 is a bit too conservative and you should change the I/sigma cutoff to 10.0

The usual empirical correction type is Ellipsoid [E], and you will have to input a value for mu*R, the absorption coefficient times the mean radius of your crystal. It is an observed fact for this kind of correction that the radius should be weighted towards the smaller half-dimensions of the crystal. You should also calculate the value of e-mu*t for your minimum crystal thickness. Compare this result to the value of Tmax which is given by the calculation. They should be about the same (i.e. 0.95 is about the same as 0.90). If they are, then the correction should be a reasonable one, and should be applied. Please do not select "print the reflections", as this makes the output file very long.

Once you have applied the absorption correction, or decided that one is not appropriate, return to the main menu and write the hkl file [F].  You will then be prompted for an output filename, the default usually suffices followed by a query about the file format.  Make sure to select the XS/SHELXS [S] format.  It is the default.  Lastly you will need to write (overwrite) the HKL file [Y].  Press <CR> to write the HKL file.

To quit XPREP press Q <CR> at the main screen (this is the default after you have follwoed this procedure).

Print the results of your analysis by typing

<...smartuse/mydir> %print filenamem.prp<CR>

Return to the Main Help Pages

2) SADABS

SADABS requires the files filename0.raw, filename1.raw, filename2.raw, and filename3.raw.

In the following example, everything except the bold words are printed by the program. (Bold-face parenthetical remarks to clarify what is going on are not part of the computer output.) The numbers are taken from an actual example, but the filenames have been changed to protect the guilty. Your numbers will be different.  For a more detailed discussion of the SADABS program and how it applies corrections to data, view the SADABS manual.

<...smartuse/mydir> % sadabs<CR>

SADABS - Bruker area detector absorption (and other) corrections

-----------------------------------------------------------------

Maximum number of reflections allowed (500000): <CR>

Enter listing filename [sad.abs]: filenames.out<CR>

(Note that the use of the substitution of "s" for "m" in the file names. This is conventional but not necessary. Just don't use the identical name to what you have been using before.)

Laue group numbers:

[1] -1................................................... [8] -3m (rhombohedral axes)

[2] 2/m (Y unique).............................. [9] -31m (Z unique)

[3] mmm ........................................... [10] -3m1 (Z unique)

[4] 4/m (Z unique)............................ [11] 6/m (Z unique)

[5] 4/mmm (Z unique)...................... [12] 6/mmm (Z unique)

[6] -3 (rhombohedral axes)............. [13] m3

[7] -3 (Z unique).............................. [14] m3m

Enter Laue group number [2]: nn<CR> --- (Pick the correct number and input it.)

Treat Friedel opposites as equivalent for parameter refinement (Y or N) ?

If you answer with Y, you may suppress the anomalous differences, but for N a higher redundancy is needed for a good correction. A chiral space group is assumed for the answer N [Y]: <CR>

(In fact, even if you have a chiral space group with anomalous scatterers, it is a good idea to use the default here. The changes induced by anomalous scattering do not follow the same symmetries as absorption and the other affects corrected for by SADABS.)

Use a centrosymmetric point group for error model and statistics [Y]: <CR>  As with merging Friedel opposites, you need a high data redundancy here, simply applying a centrosymmetric correction will generally work well in most cases.

Read reflection files written by SAINT (.raw assumed if no extension)

Enter filename (/ if no more) [ ]: filename0<CR>

Enter filename (/ if no more) [filename1.raw]: <CR>

Enter filename (/ if no more) [filename2.raw]: <CR>

Enter filename (/ if no more) [filename3.raw]: <CR>

Enter filename (/ if no more) [filename4.raw]: /<CR>(Note the "/" mark before the <CR>.)

This will read in the raw data from your integration and will allow you to begin the absorption correction process.  You will then be presented with a table of information regarding your dataset.  It will contain the number of data collected, the number of unique data, the number of redundant reflections (i.e. how many reflections were measured multiple times) and the number of reflections at a given intensity [I/sigma(I)].

It is worth noting down the number of data collected and the number of unique reflections.

You will then be presented with a series of questions regarding how to apply an absorption correction to your data.  Below is listed our recommended defaults.  The sets of variables that most influence your data (and hence are the most likely to change) are the mean(I/sigma) threshold (though using a value of 5 results in a fairly accurate analysis by the software) and the odd and even spherical harmonics.  Generally a [6,1] harmonic yields the most adequate correction.  The two values should be altered systematically and the resulting R(int) after the corrections are applied should be noted.  A gain of a few units in the last decimal place of R(int) is not a significant change and you should probably return to an earlier harmonic correction.

Enter mean(I/sigma) threshold (must be positive) [3]: 5<CR>
Highest resolution for parameter refinement [0.1]: <CR>
Factor g for initial weighting scheme  w = 1/(sigma^2(I)+(g<I>)^2), where
sigma(I) is estimated by SAINT and <I> is mean intensity [0.04]: <CR>
The following restraint esd could be increased for strong absorbers.
Restraint esd for equal consecutive scale factors [0.005]: <CR>
Suitable spherical harmonic orders are 4,1 for weak absorption and 8,5 for
strong.  Highest even order for spherical harmonics (0,2,4,6 or 8) [6]: <CR>
Highest odd order for spherical harmonics (0,1,3,5 or 7) [3]: 1<CR>
Allow for crystal decomposition by B-value refinement [N]: <CR>
Number of refinement cycles [15]: <CR>

nnnn Reflections employed for parameter determination

Effective data to parameter ratio =x.yy

R(int) = 0.nnn before parameter refinement

<List of R(int)'s across each of the succesive cycles>

R(int) = 0.mmm  (selected reflections only, after parameter refinement).  (Another useful number to note down, the final R(int) of the data set after the absorption correction has been applied.)

(At this point you can either do as follows and write out the processed data, or you can test other parameters by Repeating the analysis. If your crystal shape is unusual or you just want to test the results, try increasing the number of even coefficients.)

 Repeat parameter refinement (R) or accept (A) [A]: <CR> [or R<CR> (to repeat the parameter refinement)]

You will now begin the second portion of data analysis and correction;
 
PART 2 - Reject outliers and establish error model

 High resolution limit [0.1]:<CR>  If your data were weaker than you expected when you integrated the data (for example if you integrated the data to 0.8 A and after analysis of the SAINT output it was revealed that the real resolution was approximately 0.95 A) you could apply the high angle cut-off here if you so choose.

|I-<I>|/su ratio for rejection [4.0]: 10<CR>  This determines the upper limit for seriously erroneous data.  Setting this value to 10 allows more "outliers" to be included in the dataset, or in another way to look at it, more of the original data to be included in the final dataset.

g-value: (accept the default at this stage).

You will then be given a new g-value based on the software's own calculations.  You should accept the default.

A table detailing each block of data, the R(int), calculated transmission factors, goodness of Fit (of the data), the number of data in each block and the number of strong [I>2sigma(I)] data.  For example:

  Run 2-theta  R(int)  Incid. factors  Diffr. factors    K     Total I>2sig(I)
    1  -25.0  0.0621   1.323 - 1.464   0.911 - 1.067   0.767    9968    5254
    2  -25.0  0.0684   1.379 - 1.524   0.910 - 1.061   0.812    7126    3664
    3  -25.0  0.0675   0.435 - 0.502   0.936 - 1.060   0.786    3718    1851
    4  -25.0  0.0605   0.376 - 0.393   0.942 - 1.054   0.734     916     511

[[NOTE: The numbers for the last two runs are anomalous, and it would probably be a good idea in this case to select the P option for the next question and try a new value for the even-order spherical harmonics. The range of values for the Incident factors and the Diffr. factors should be similar, not off by a factor of two. If you have high absorption and a crystal of extreme shape this may not be the case.]]

You can then be asked if you wish to repeat the absorption correction process/parmeter refinement (P), repeat the rejection criteria process (R) or to continue (A to accept).

When you choose to accept the data model you have established, you will begin the final section of the program, writing out relevant data.

PART 3 - Output Postscript diagnostics and corrected data

 Write Postscript diagnostic file (Y or N) [Y]:  You choose.  This function creates a postscript image that can be printed out.  It contains information about the quality of the data across the entire data collection.  Generally not useful unless you suspect something seriously wrong with your data (for example if the R(int) of one of the blocks was significantly different from the rest)

Repeat (R), write .hkl file (W) or quit (Q) [W]: <CR>

Enter name of output .hkl file [sad.hkl]: filenames.hkl<CR>  (NOTE: an "s" is used here to indicate that the data were processed using sadabs, you can use any name, but simply be careful not to over-write any other data).

 Mu*r of equivalent sphere for additional spherical absorption correction.
 Enter <CR> if none: n.nn<CR> (Enter the value that you calculate based on mu from XPREP and the average radius of crystal)

 Lambda/2 correction factor (0 if none, e.g. for MWPC!) [0.0015]: <CR>  This correction can be modified if you collect data on a different diffractometer (0.0015 is the correct value for the laboratory instruments). For example, data collected at beamline 11.3.1 the Advanced Light Source should have a lambda/2 correction factor of 0 (zero) applied.

xxxx Corrected reflections written to file filenames.hkl

Minimum/Maximum effective transmission: y.yyyyy

Repeat (R), write .hkl file (W) or quit (Q) [Q]: <CR> (This returns you to the system prompt.)

<...smartuse/mydir>%

Print the results of your analysis by typing

<...smartuse/mydir> %print filenames.out<CR>

Now back up your frames to CD, following the data archiving procedures.

Now move on to creating an account on BlueCHEX and solving your dataset.

(Return to the Help Files page)